RE: OWL Full proposal (sort of) - addressing my Action

Alan Ruttenberg wrote:

>What I think I'm hearing is a proposal for an *informative* section  
>in the documents giving a (possibly incomplete) definition of OWL, in  
>the interest of making it more understandable.

I didn't really expect it to be possible to get a whole rule set for
OWL-Full, nor did I believe that this is necessary. Instead, as you say
above, I suggested in my earlier post to define some "good selection". But
now that the talk is about RDFS-3.0 (or OWL-Prime, or whatever it will be
called), it looks to me that it might make some sense to try to create a
rule scheme for this language.

This would then probably need the following steps to be performed:

  (1) Define the vocabulary of RDFS-3.0 (probably some variation of the
current proposal [1]).

  (2) Select a subset of the OWL-Full semantic conditions, which covers the
vocabulary of RDFS-3.0. 

  (3) Select a set of triple rules, which is sound and complete w.r.t. these
chosen semantic conditions.

By "sound and complete" I mean the following:

  For each two RDF graphs G and G':
  G entails G' w.r.t. the model-theoretic semantic conditions of RDFS-3.0
  if and only if
  there exists some finite sequence of triple rule applications leading from
G to G'.

Such a rule set would then at the same time

  (a) be a "co"-specification of RDFS-3.0 semantics, in addition to the
model-theoretic specification; 
  
  (b) be a hint to reasoner implementors, without saying too much about how
the implementation should be performed; 

  (c) meet educational purposes by giving a restricted and simplified view
on how OWL-Full "works";

  (d) copycat the way how RDFS semantics is introduced in the RDF semantics
spec: by giving an additional informative section which presents a set of
triple rules [2].

I hope that it is realistic to make this happen. A very first glimpse on the
terHorst paper, which Ivan refered to recently, gives me some hope. But I
need to find the time to read it in detail, before I can make any further
assumptions.

Cheers,
Michael

[1]
<http://lists.w3.org/Archives/Public/public-owl-wg/2008Jan/att-0308/00-part>
[2] <http://www.w3.org/TR/rdf-mt/#rules>

>
>-Alan
>
>On Feb 14, 2008, at 10:25 AM, Jeremy Carroll wrote:
>
>>
>> Ivan Herman wrote:
>>> I *always* use those entailement rules to explain,
>>
>> I think OWL gets too complicated to express only by means of rules.
>>
>> I am trying to make a formal point, that I am sure somewhat else  
>> could make better.
>>
>> Essential rules work for RDF, RDFS, and even pD* because if you  
>> apply all the rules until they can't apply anymore (and take  
>> appropriate steps with certain problems) you can end up with a  
>> workable piece of code (for example Jena rules).
>>
>> But this approach fails if taken to the limit.
>>
>> I guess it would be possible to have a set of rules that was not  
>> practical in that way (that the closure is badly infinite, i.e.  
>> infinite in ways which you can't work around), which did articulate  
>> the semantics of OWL ....
>>
>>
>> Jeremy
>>
>>
>
>

--
Dipl.-Inform. Michael Schneider
FZI Forschungszentrum Informatik Karlsruhe
Abtl. Information Process Engineering (IPE)
Tel  : +49-721-9654-726
Fax  : +49-721-9654-727
Email: Michael.Schneider@fzi.de
Web  : http://www.fzi.de/ipe/eng/mitarbeiter.php?id=555

FZI Forschungszentrum Informatik an der Universität Karlsruhe
Haid-und-Neu-Str. 10-14, D-76131 Karlsruhe
Tel.: +49-721-9654-0, Fax: +49-721-9654-959
Stiftung des bürgerlichen Rechts
Az: 14-0563.1 Regierungspräsidium Karlsruhe
Vorstand: Rüdiger Dillmann, Michael Flor, Jivka Ovtcharova, Rudi Studer
Vorsitzender des Kuratoriums: Ministerialdirigent Günther Leßnerkraus